Forum Discussion
Graph http 449 throttled
We are experiencing a lot of Microsoft Graph trolling errors from some of our applications that are hosted in Azure Web App Services and other third party such as Front App to name one. I am trying to find an approach or strategy to figure out and narrow down what may be causing so many of these events. Whether I can narrow down by application or process, I am not sure yet. We enforce MFA on all our users, but of course, on Azure Enterprise Applications, we don't, which are used extensively in our ecosystem of apps. Any help is much appreciated here.
1 Reply
Would suggest locating the issues first:
• Correlation headers: Capture client-request-id, request-id, and x-ms-ags-diagnostic in every failed call.
• Retry hints: Log Retry-After and any throttling headers. Respect the longest hint across chained calls.
• Request identity: Record whether calls are delegated (user) or application (SPN), plus appId, tenant, and scopes.
• Endpoint signature: Log the full URI (minus PII), especially the resource and query options ($select, $filter, $top, etc.).
• Concurrency context: Track queue depth, parallel workers, and per-endpoint in-flight counts.
Dashboards to build (App Insights or equivalent)
• Errors by appId: 429/449 counts grouped by application, credential type (delegated vs app), and environment slot.
• Errors by endpoint: Identify hot endpoints (e.g., /users, /messages, /drive/root/children, /sites/{id}/lists/...).
• Temporal patterns: Heatmap of throttles by minute/hour; correlate with scheduled jobs and vendor integrations.
• Burst detection: Distribution of request rates and concurrent calls per app; flag spikes above your soft caps.
This quickly tells you if a single app or a specific workload (often SharePoint, Mail, or Directory) is the problem.
